| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
0 Harvard University Herbaria, 22 Divinity Avenue, Cambridge, Massachusetts 02138 USA
Received for publication July 27, 1999. Accepted for publication December 16, 1999.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: breeding system evolution • coevolution • dioecy • Ficus • Moraceae • phylogeny • pollination
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). These species of trees, shrubs, climbers, and hemiepiphytic stranglers are recognized by a specialized inflorescence and pollination syndrome (Janzen, 1979
; Berg, 1990b
). Resembling a fleshy fruit, the fig is an enlarged receptacle enclosing many unisexual flowers that are accessible only by a tightly bract-filled opening or ostiole. The closed inflorescence, or syconium, protects the flowers against most parasites except for diminutive insects capable of entering through the opening (Berg, 1990a
). The interior of the inflorescence is the location of an obligate mutualism with pollinating seed predators, fig wasps in the family Agaonidae of parasitic Hymenoptera (Chalcidoidea). Interactions between figs and fig wasps are among the best known examples of reproductive interdependence between plants and their pollinators (Bronstein, 1992
). In addition, fig wasps are specialized to the extent that unique pollinator species are associated with most fig species (Ramirez, 1970
; Wiebes, 1979
; but see Rasplus, 1994, 1996
; Kerdelhue, Hochberg, and Rasplus, 1997
).
The intertwined life cycles of figs and pollinators, together with their extreme specificity, are the basis for speculation on the nature and extent of coevolution involved (Ramirez, 1974
; Wiebes, 1979
). Functionally dioecious figs are of special interest due to evolutionary conflicts with pollinators regarding the exploitation of seed resources (Grafen and Godfray, 1991
). Until the present, our knowledge of fig breeding system evolution has been shaped by taxonomy (Corner, 1985
), anatomy (Beck and Lord, 1988
; Verkerke, 1989
), ecology (Galil, 1973
; Kjellberg et al., 1987
; Corlett, 1993
; Weiblen, Spencer, and Flick, 1995
; Patel and McKey, 1998
), and pollinator behavior (Hossaert-McKey, Gibernau, and Frey, 1994
; Ware and Compton, 1994
). However, phylogenetic studies are few, and the relationships of the functionally dioecious figs have not been examined in detail (Yokoyama, 1995
; Herre et al., 1996
).
Morphologically, figs are monoecious or gynodioecious according to the representation of the unisexual florets within the syconium (Figs. 1–3). The gynodioecious species are functionally dioecious, with the separation of sexual function resulting from the interaction of pollinator wasps with florets in two types of figs on separate plants (Weiblen, Spencer, and Flick, 1995
). Inside the protogynous syconia, female fig wasps deliver pollen to the pistillate florets while laying their eggs in a fraction of fig ovaries. The fate of the ovaries in these species is influenced by the interaction of pollinator ovipositors with dimorphic pistillate florets in the two types of figs (Ganeshaiah et al., 1995
). Long-styled florets in seed figs (Fig. 1) have ovaries that are fertilized but are inaccessible to pollinator ovipositors (Galil, 1973
). On the other hand, short-styled florets in gall figs (Fig. 2) enable pollinators to deposit their eggs in proximity to fig ovules (Verkerke, 1989
). Wasp larvae occupy the ovaries and feed on endosperm. Later in development, staminate florets in gall figs release pollen during the eclosion and mating of the adult wasps. The wingless males then chew an exit from the fig and the winged, pollen-bearing females escape in search of receptive figs in which to complete their life cycle.
|
). The genetics of sex determination in functionally dioecious Ficus is known from crossing studies in the edible fig, F. carica L. (Storey, 1975
). A pair of linked loci affecting style lengths and the abortion of staminate florets is responsible for gynodioecious morphology. A stable ratio of progeny results from crosses between heterozygous gall figs (GgAa) and homozygous seed figs (ggaa); G is dominant for short-styled pistillate florets, while g is recessive for long-styled florets, and A is dominant for the development of staminate florets, while a is recessive for the abortion of staminate florets. The evolutionary sequence leading to the origin or loss of this unique sex-determining mechanism is unknown (Valdeyron and Lloyd, 1979
), although selective pressures such as environmental seasonality and the impact of non-pollinating wasps have been proposed (Kerdelhue and Rasplus, 1996
).
A phylogenetic analysis of the functionally dioecious figs and their relatives was undertaken to examine breeding system evolution, pollinator relations, and Ficus classification. In the last century, Ficus was split into several genera (Gasparrini, 1844
; Miquel, 1862
) that became the basis for a subgeneric classification after the genus was reunited (Miquel, 1867a, b
). Miquel classified the functionally dioecious species in four subgenera based on floral characters. Almost a century later, Corner (1965)
united functionally dioecious figs under subg. Ficus in his reclassification of the genus (Table 1). Ficus can be found in all three tropical regions, but the majority of species occur in Asia, Malesia, and Australia. Functionally dioecious subg. Ficus is restricted to the Paleotropics, and Malesia is the center of diversity in terms of species richness and endemism. For example, functionally dioecious figs comprise an estimated 343 species out of the 503 species in the region (68%; Berg, 1989
). In addition, five of the eight functionally dioecious sections are centered in Indo-Australia (Table 1). Species within this region were the focus of the study.
|
). As a supplement to ITS sequences, morphological characters for Ficus were analyzed separately and in combination. Whether or not to combine morphological and molecular data sets in a single analysis has been a subject of considerable debate in the systematic literature (Bull et al., 1993
; de Queiroz, Donoghue, and Kim, 1995
). Assuming that different data sets share the same phylogenetic history, systematists have argued that inferences based on all the available data are more likely to be correct than inferences based on a subset of the data (Kluge, 1989
; Barrett, Donoghue, and Sober, 1991
). However, conflict between data sets can result from systematic error, rate heterogeneity, or from data sets not sharing the same history.
Separate analyses have the advantage of highlighting points of conflict, but if incongruence is due to random errors in phylogeny estimation, then a combined analysis may provide the best estimate of phylogeny (de Queiroz, Donoghue, and Kim, 1995
). A conditional approach favors combined analyses in the event of weak incongruence while favoring separate analyses in the event of strong incongruence. Considerations on separate vs. combined analysis of ITS and morphological data sets for Ficus were explored using statistical tests. Herre et al. (1996)
suggested that Ficus morphology may yield incorrect estimates of phylogeny because of convergent evolution in reproductive traits; however previous studies did not specifically test this proposition. With regard to the question of breeding system evolution, the issue of including characters of interest in phylogeny reconstruction was examined using sensitivity analysis (de Queiroz, 1996
; Donoghue and Ackerly, 1996
).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
) and morphology (Berg, 1989
) suggests that the neotropical Ficus sect. Pharmacosycea is a sister group to the rest of Ficus. Two representatives of sect. Pharmacosycea were designated as outgroups. Sampling of the monoecious subgenera included 15 species representing the sections Oreosycea, Urostigma, Conosycea, Malvanthera, Americana, and Sycomorus. In addition, 29 species comprising 8% of subg. Ficus and at least two representatives of each functionally dioecious section were sampled, excluding monotypic sect. Sinosycidium.
|
was modified to avoid problems associated with DNA isolation from leaves containing latex. Leaves were ground in liquid N2 and incubated at 60°C in a 400-µL solution of 2X CTAB buffer with 4% polyvinyl pyrrolidone (molecular weight 40 000) and 0.8 µL of ß-mercapto-ethanol. After 1 h, samples were centrifuged for 5 min, and the aqueous supernatant was twice extracted with 400 µL of phenol:chloroform:isoamyl alcohol (25:24:1). The supernatant was extracted a third time with chloroform:isoamyl alcohol (24:1). DNA extracts were cleaned with a GENECLEAN II® kit (BIO 101, Carlsbad, California, USA), serially diluted, and amplified with a PCR reagent system (Gibco BRL, Rockville, Maryland, USA).
Primers ITS4 and ITS5 (White et al., 1990
) were used for amplification of the region including the two internal transcribed spacers and the 5.8S subunit of nuclear ribosomal DNA. The thermal conditions for amplification included: (A) denaturation at 96°C (2 min); (B) two cycles of denaturation at 94°C (30 s), annealing at 40°C (30 s), and extension at 72°C (60 s); (C) 35 cycles as in (B) but with annealing at 55°C (30 s); and (D) final extension at 72°C (4 min). Polymerase chain reaction (PCR) products were quantified on 0.4% agarose gels using a Low DNA MassTM ladder (Gibco BRL, Rockville, Maryland, USA) and single bands were purified with a QIAquickTM PCR purification kit (QIAGEN, Valencia, California, USA). PCR products were cycle sequenced in both directions using primers ITS2, ITS3, ITS4, and ITS5 (White et al., 1990
). ITS2 and ITS3 sequencing primers were redesigned for Ficus (5'-GCATCGATGAAGAACGTAGC-3' and 5'-GGAAGGAGAAGTCGTAACAAGG-3', respectively). Sequences were collected using Long RangerTM polyacrylamide gels (FMC Bioproducts, Rockland, Maine, USA) and a 377 PRISMTM sequencer with DNA Sequencing Analysis software version 2.1.1 (PE Biosystems, Foster City, California, USA). Chromatograms were edited with SequencerTM software (Gene Codes, Ann Arbor, Michigan, USA) and aligned manually. The aligned sequences are deposited in TreeBASE (http://www.herbaria.harvard.edu/treebase/index.html). Thirty-three ambiguous positions corresponding to 4.3% of the aligned sequences were excluded from analysis. Thirty-four gaps remained in the aligned sequence following the exclusion of these ambiguous sites. Nineteen autapomorphic indels were treated as missing data. The presence or absence of 15 remaining indels was coded in a supplemental set of characters, and all indel positions were excluded from analyses of the aligned sequences.
Molecular cloning examined heterogeneity among ITS paralogues in functionally dioecious figs. ITS heterogeneity within species was explored because the inclusion of divergent paralogues and pseudogenes in phylogenetic analysis has the potential to yield inaccurate estimates of species phylogeny (Buckler, Ippolito, and Holtsford, 1997
). PCR products from four species were cloned and sequenced for comparison with the results of direct sequencing. In addition, ten ITS clones each from F. nodosa and F. variegata were sequenced to look for the presence of heterologous ITS copies within species. ITS PCR products were ligated and transformed using the pGEM®-T Easy Vector System (Promega Corporation, Madison, Wisconsin, USA). Transformed cells were screened with ampicillin, and recombinant plasmid DNA was isolated using the Wizard® Plus Miniprep DNA purification system (Promega Corporation, Madison, Wisconsin, USA).
Sixty-four discrete morphological characters were selected from the taxonomic literature (Corner, 1933, 1955, 1958, 1960a, b, 1961, 1965, 1967, 1969, 1970a, b, 1976, 1978
) and by examination of living plants and more than 800 herbarium collections. Sixty-one characters with two to five states were potentially informative in phylogenetic analysis (Appendix).
Phylogenetic analyses were performed with PAUP* version 4.0b2 for Power Macintosh computers (Swofford, 1998
). Under the optimality criterion of parsimony, heuristic searches were conducted according to PAUP* default settings, except that 1000 random addition sequence replicates were used with MAXTREES set to increase without limit. All characters were unordered and weighted equally. Uninformative characters were excluded from all analyses. Bootstrap resampling (Felsenstein, 1985
) and decay analyses (Bremer, 1988
; Donoghue et al., 1992
) were used to estimate clade robustness. Bootstrapping involved heuristic searches with 10 000 replicates and a random addition sequence with N = 1. In the case of ITS and combined analyses, the option to save multiple equally parsimonious trees per replicate was disabled to reduce the search times on Power Macintosh 7300 and Macintosh G3 computers. Decay analyses were performed using the program Autodecay version 2.9.5 (T. Eriksson) with ten random addition sequence replicates per heuristic search.
Congruence between morphological and molecular data sets was explored with the incongruence length difference test (ILD; Farris et al., 1994
) and Templeton tests (Templeton, 1993
; Larson, 1994
) as implemented in PAUP*. Each data set was analyzed to find the most parsimonious trees compatible with constraint trees from the rival data set. For example, morphological data were analyzed to find the shortest trees compatible with the shortest trees from a separate analysis of the ITS data. Constraint trees from the rival data set included the strict consensus tree, bootstrap consensus trees (50, 70, and 90%), and a most parsimonious tree selected at random. Most parsimonious trees from the constrained and unconstrained searches were selected at random and compared using a nonparametric sum of signed ranks test. In addition to statistical measures of conflict, comparisons of bootstrap values between conflicting nodes in the separate analyses were used to identify points of weak and strong incongruence between morphological and ITS trees.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
1%) in F. nodosa and F. variegata, respectively. The location of nucleotide differences among ten clones from each species was scattered such that clones could not be grouped below the level of species. This kind of heterogeneity was suggestive of errors by DNA polymerase during cycle sequencing reactions possibly induced by high GC content in the ITS region. Overall, the results of cloning and direct sequencing suggest that heterogeneity in the ITS region did not pose a major problem for Ficus phylogeny reconstruction. Manually aligned ITS sequences for 46 species were 761 bp in length including 33 positions with ambiguous alignment. Parsimony analyses of ITS alone were based on 643 bp excluding the ambiguous positions and indel positions coded as supplemental characters. One hundred and sixty-five nucleotide positions (25.6%) were potentially informative. In addition, 15 out of 35 indels were potentially informative. Analysis of the 180 characters combined found a single island of 208 most parsimonious trees of 453 steps (consistency index or CI = 0.55). The strict consensus was congruent with the bootstrap consensus at 29 of 31 nodes with >50% support (Fig. 4). Two clades with bootstrap values <60% did not appear in the strict consensus but are shown in Fig. 4. A clade with F. pungens as sister to subsect. Sycocarpus and a clade with F. septica as sister to the rest of subsect. Sycocarpus were compatible with 158 and 50 out of 216 most parsimonious trees, respectively.
|
), the neotropical and paleotropical sections of subg. Pharmacosycea did not form a clade (Fig. 4), and there was marginal support from ITS for the paraphyly of sect. Oreosycea. Monoecious subg. Urostigma was not monophyletic due to the position of sect. Urostigma as sister to a functionally dioecious clade, but support for this relationship was weak. Subgenus Ficus was polyphyletic and divided into two highly supported clades. One functionally dioecious clade included the well-supported and monophyletic sects. Ficus, Kalosyce, Rhizocladus, and Sycidium, excluding F. pungens. The other clade included functionally dioecious sects. Adenosperma, Neomorphe, Sycocarpus, F. pungens, and monoecious subg. Sycomorus. Relationships within this clade were not well resolved, although monophyly of sect. Adenosperma, subsect. Sycocarpus, and subg. Sycomorus were each highly supported. In addition, the derivation of monoecious subg. Sycomorus within functionally dioecious subg. Ficus received strong bootstrap support.
Morphology
The morphological analysis yielded six most parsimonious trees of 339 steps (CI = 0.47). The strict consensus was congruent with the bootstrap consensus at 20 out of 21 nodes with >50% support. (Fig. 5). A clade representing neotropical sect. Pharmacosycea with 57%, shown in Fig. 5, was not present in the strict consensus due to the position of F. albipila as sister to F. insipida in the most parsimonious trees. Morphological analysis indicated that subg. Ficus was not monophyletic and that monoecious subg. Sycomorus was derived within a paraphyletic sect. Neomorphe. The functionally dioecious figs including subg. Sycomorus were sister to monoecious subg. Urostigma. These three subgenera were derived within a paraphyletic subg. Pharmacosycea. However, morphological support for subgeneric relationships was relatively weak, as indicated by low bootstrap values at deep nodes compared to shallow nodes. Monoecious subg. Urostigma had a bootstrap value of 64%, but support for the paraphyly of sect. Oreosycea was lacking. The monophyly of monoecious sects. Conosycea, Malvanthera, and Urostigma was upheld in the morphological analysis, in contrast to dioecious sects. Sycidium, Sycocarpus, and Neomorphe, which were polyphyletic or paraphyletic. However, highly supported clades of functionally dioecious figs included sects. Adenosperma, Ficus, Kalosyce, and Rhizocladus.
|
).
Results of Templeton tests of incongruence are summarized in Table 3. ITS sequence data strongly rejected the shortest morphological trees. Similarly, the morphological data rejected the shortest ITS trees. However, taking into account clade robustness in the separate analyses had a strong impact on the test results. For example, ITS sequences marginally rejected the morphology-based 50% bootstrap consensus and morphological data significantly rejected the ITS 50% bootstrap consensus, but neither ITS nor morphological data sets rejected their rival 70 or 90% bootstrap consensus trees. Therefore, statistically significant conflict between ITS and morphology was limited to weakly supported nodes in the separate analyses, and it appeared that weak incongruence due to systematic error or rate heterogeneity could account for the significant ILD (Bull et al., 1993
).
|
|
|
), neotropical sect. Pharmacosycea and paleotropical sect. Oreosycea did not form a clade. Within monoecious subg. Urostigma, there was strong support for the monophyly of the Indo-Australian sects. Conosycea, Malvanthera, and Urostigma. Two strongly supported clades containing functionally dioecious figs were also recovered in the combined analysis, but it was not entirely clear whether these were sister groups (Fig. 6). The first of these clades was functionally dioecious in its entirety and included sects. Ficus, Kalosyce, Rhizocladus, and Sycidium (excluding F. pungens). Bootstrap and decay values strongly supported the monophyly of each of these sections and the exclusion of F. pungens from Sycidium. Relationships within Sycidium were resolved but mostly not supported by high bootstrap values. Sections Kalosyce and Rhizocladus were strongly supported sister groups, and this clade was sister to sect. Ficus. A second major clade of functionally dioecious figs, including F. pungens, sects. Adenosperma, Neomorphe, and Sycocarpus, with monoecious subg. Sycomorus, had high bootstrap support (89%). However, basal relationships within this clade were not well resolved in the combined analysis. Although sect. Sycocarpus was clearly not monophyletic, relationships within the section were mostly unresolved. Section Neomorphe was not monophyletic due to the highly supported relationship of F. semivestita to sect. Adenosperma. Members of sect. Neomorphe (excluding F. semivestita) belong to a well-supported clade including monoecious subg. Sycomorus. The sister relationship between functionally dioecious F. itoana and monoecious F. microdictya also received strong support in the combined analysis.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Campbell et al., 1997
) were not detected in Ficus. Divergent ITS paralogues appear to be common in lineages having a history of hybridization and polyploidy (Buckler, Ippolito, and Holtsford, 1997
). However, natural hybridization and polyploidy are rare in Ficus (Storey, 1975
). Cytology has been examined in over 100 species and most are diploid (2n = 26), with exceptions such as a sterile triploid (3n = 39) cultivar of F. elastica Roxb. (e.g., Condit, 1964
).
Although ITS was phylogenetically informative, the ability to resolve relationships within Ficus was limited. Manual alignment of ITS sequences within Ficus was straightforward, but sequences were highly diverged compared to other Moraceae and alignment across other genera in the family was complicated by the presence of overlapping indels. This required that tree rooting be based on additional molecular, morphological, and paleontological data (Brues, 1910
; Berg, 1989
; Herre et al., 1996
). Low levels of sequence variability among closely related species also limited the utility of ITS for resolving phylogenetic relationships within most sections. For example, nearly identical ITS sequences were obtained from closely related taxa, such as F. odoardi and F. bauerlenii (sect. Rhizocladus) or F. bernaysii and F. hispidioides (subsect. Sycocarpus). Additional gene regions are needed to corroborate the results based on ITS and morphology.
Morphology and functionally dioecious fig phylogeny
The use of morphological characters in phylogeny reconstruction is worthy of consideration when examining the relationships of the functionally dioecious figs. It has been suggested that convergence in functional traits might bias our conclusions about phylogenetic relationships (Wiebes, 1994b
; Herre et al., 1996
). Examples of correlated traits include: (1) the construction of the ostiole in figs and pollinator head shape (van Noort and Compton, 1996
); (2) the distribution of staminate florets in figs and pollinator behaviors or structures associated with pollen collection (Ramirez, 1978
); and (3) fig breeding system and pollinator ovipositor lengths (Ramirez, 1980
). However, it can be argued that most characters, whether morphological or molecular, exhibit homoplasy, and this is not sufficient for their exclusion from phylogenetic analysis (Donoghue and Sanderson, 1992
). In this study, ITS sequences and morphology showed similar levels of homoplasy in Ficus (CI = 0.54 and CI = 0.46, respectively). Morphological features (Appendix) also provide useful touchstones for recognizing clades within Ficus, and lists of apomorphies are described in Weiblen (1999)
.
Tests of incongruence
Significant incongruence was detected between the morphological and ITS data sets by both ILD and Templeton tests (Table 3). However, it is important to ask whether statistically significant conflict, as measured by these tests, should be interpreted as evidence of different phylogenetic histories or rate heterogeneity (Bull et al., 1993
). Incongruence may also be due to systematic error, and Templeton tests are potentially informative in this regard because they can take into account levels of support for rival clades in separate analyses. Templeton test results were highly sensitive to the choice of rival constraint trees. If weakly supported clades were included in rival constraint trees, the data significantly rejected the null hypothesis that random errors in phylogeny estimation account for length differences in rival trees. However, the null hypothesis was not rejected if only strongly supported clades were included. This was true for both ITS and morphology when rival constraint trees were limited to clades with >70% and >90% bootstrap support. The arbitrary level of bootstrap support considered "strong" seemed reasonable based on empirical studies of phylogenetic accuracy (Hillis and Bull, 1993
).
The overall results provided little evidence of strong incongruence between data sets, although several instances of local incongruence deserve further consideration. For example, ITS and morphological analyses differed with regard to the monophyly of subg. Urostigma. Mophological and combined analyses placed subg. Urostigma in a clade with 65 and 64% bootstrap support, respectively (Figs. 5–6). On the other hand, ITS alone placed sect. Urostigma as the sister to a functionally dioecious clade with 59% support (Fig. 4). Decay analysis for ITS indicated that five additional steps were required to contradict the separation of sect. Urostigma from the rest of subg. Urostigma. However, reciprocal Templeton tests of local incongruence were not significant (P = 0.20 and P = 0.37 for ITS and morphology, respectively). Another local conflict involved a clade including F. botryocarpa, F. hispidioides, and F. septica with 58% bootstrap support in the morphological analysis vs. a rival clade including F. bernaysii, F. botryocarpa, and F. hispidioides with 54% bootstrap support in the ITS analysis. Both clades were present in the combined most parsimonious trees, and support from separate analyses was relatively weak (<60%). In future exploration of local conflicts, it might be possible to minimize the effects of systematic error through modification of character weighting under parsimony or alteration of rate parameters under maximum likelihood (de Queiroz, Donoghue, and Kim, 1995
; Huelsenbeck, Bull, and Cunningham, 1996).
Comparisons of bootstrap values as measures of relative clade support indicated that the results of combined analyses were better supported than either separate analysis. Compared to ITS, bootstrap values for 15 nodes increased in the combined analysis, while support for six nodes decreased. Compared to the separate morphological analysis, bootstrap support for 21 nodes increased in the combined analysis, while none decreased. In the absence of evidence for strong incongruence, the combined data provided the best supported estimate of functionally dioecious fig phylogeny. Similar conclusions have been reached in studies of other plant groups (Manos, 1997
; Kelley, 1998
; but see Mason-Gamer and Kellogg, 1996
). Classification, breeding system evolution, and associations with pollinators will be discussed in terms of the combined analysis (Figs. 6–8). However, inferences from the combined phylogeny should be regarded as preliminary until corroborated by analyses of additional genes and taxa.
Classification of functionally dioecious figs
Phylogenetic analysis of ITS and morphological characters sheds light on the traditional classification of functionally dioecious figs (Corner, 1965
) and the revised classification of Ficus based on pollinator taxonomy (Ramirez, 1977
; Berg, 1989
). Although some groups appear to be monophyletic, the functionally dioecious figs are not (Fig. 6). In general, Corner's subgeneric classification (Table 1) is not supported by the result of ITS, morphology, or combined analyses (Figs. 4–6). Pharmacosycea is not monophyletic, and, in spite of morphological similarity, neotropical sect. Pharmacosycea and paleotropical sect. Oreosycea do not form a clade. Based on the combined analysis, Urostigma may be monophyletic, but separate analyses disagree in this regard. There is no tree supporting the monophyly of sect. Oreosycea although ITS and combined analyses support the existence of two major clades of functionally dioecious figs derived within paraphyletic sect. Oreosycea. Whether the functionally dioecious clades are sister groups, however, is unclear from the combined analyses.
Separate and combined analyses indicate that monoecious Sycomorus is monophyletic and nested in a clade of functionally dioecious Ficus. The close relationship of monoecious Sycomorus and functionally dioecious Ficus was first noted by Miquel (1867)
and again by King (1887), but Corner (1960b)
kept Sycomorus separate on the sole basis of breeding system. Ramirez (1977)
proposed a revised classification based on pollinator taxonomy that included sects. Adenosperma, Neomorphe, Sycocarpus, Sycomorus, and all Ceratosolen-pollinated Sycidium in subg. Sycomorus. A Ceratosolen-pollinated clade was indeed recovered in the phylogenetic analysis, although basal relationships within it were not well resolved. Phylogenetic analyses support the view of Berg (1989)
that sect. Adenosperma and sect. Sycocarpus are closely related. Also in agreement with Berg (1989)
, a clade including sect. Neomorphe, subsect. Sycocarpus, and subg. Sycomorus is well supported in the ITS and combined analyses.
One entirely functionally dioecious clade has no parallel in Corner's classification, but instead corresponds to subg. Ficus sensu Ramirez (1977)
, including sects. Ficus, Kalosyce, Rhizocladus and Sycidum, excluding all Ceratosolen-pollinated species. Within this clade there are two distinct groups that were recognized by Berg (1989)
, one including sect. Sycidium and the other sects. Ficus, Kalosyce plus Rhizocladus. The combined phylogenetic analysis also strongly supports the monophyly of each of these sections, excluding all Ceratosolen-pollinated Sycidium, which Ramirez (1977)
transferred into his revised subg. Sycomorus. In general, the combined phylogenetic analysis supports the alternative classification of Ramirez (1977)
.
Breeding system evolution
Phylogenetic analysis indicates one or two independent origins of functional dioecy from monoecy in Ficus, depending on which island of most parsimonious trees is examined (Figs. 7–8). The first island included seven equally parsimonious trees with monoecious F. albipila as sister to a clade containing functionally dioecious subg. Ficus plus monoecious subg. Sycomorus (Fig. 7). The second tree island showed monoecious subg. Urostigma as sister to one clade of functionally dioecious figs and F. albipila as sister to another functionally dioecious clade (Fig. 8). Both separate and combined analyses unequivocally suggested two reversals from functional dioecy to monoecy within one of the functionally dioecious lineages.
It has been argued that the characters of interest should be excluded from phylogenetic analysis in order to avoid circularity and bias in studies of character evolution. Indeed, morphological characters are sometimes excluded from phylogenetic analyses on the grounds that convergence in function can lead to inaccuracy and such arguments are often the basis for preferential use of independent molecular data (Herre et al., 1996
; Van Noort and Compton, 1996
). However, morphological and molecular data may show similar levels of homoplasy (Donoghue and Sanderson, 1992
), and molecular data may also show convergence (Naylor and Brown, 1998
). Furthermore, it is possible that excluding the characters of interest can yield biased or inaccurate results (Luckow and Bruneau, 1997
), while sensitivity analyses can examine the effect of excluding characters on inferences of character evolution (de Queiroz, 1996
; Donoghue and Ackerly, 1996
).
Exclusion of breeding system from the combined analysis resulted in one most parsimonious tree similar to the second island from the analysis based on all characters (Fig. 8). The tree indicates that monoecy is ancestral, that functional dioecy has evolved twice, and that two reversals to monoecy have occurred in one of the functionally dioecious lineages. Exclusion of additional characters possibly linked to breeding system, such as the presence of staminodes in seed figs, setose long-styled florets, and funnelform stigmas in short-styled florets, resulted in the same topology. Exclusion of all morphological characters on the grounds that they are not independent of breeding system also results in two gains of functional dioecy and two losses. Therefore, the inclusion or exclusion of morphological characters did not have an impact on the inferences of breeding system evolution, although whether functional dioecy evolved once or twice was unclear from the combined analysis. It was not possible to obtain recent collections of monoecious F. pritchardii Seem., which may represent a third reversal to monoecy within a functionally dioecious lineage. Overall, morphology and pollinator associations suggest its placement with F. pungens and subsect. Sycocarpus (Wiebes, 1963
).
Corner (1965)
delimited subgenera on the basis of breeding system, but this character appears to be more homoplasious than morphology in general. The consistency index of breeding system ranged from 0.25 to 0.33, depending on the tree island from the combined analysis, compared to 0.46 for morphology overall. The weight that Corner placed on breeding system led to splitting monoecious Sycomorus from functionally dioecious Neomorphe in spite of shared features including cauliflory and buttresses. Similarly, monoecious F. microdictya was at one time classified with sect. Oreosycea (Corner, 1965
). However, Corner (1962, 1970b
) recognized the close relationship of monoecious F. microdictya to functionally dioecious F. itoana, and phylogenetic analysis shows the monoecy of F. microdictya to be a reversal within a functionally dioecious lineage. Why, then, did Corner divide the figs primarily according to breeding system? Although shifts in breeding system are widespread in flowering plants, taxonomists have often recognized genera and subgenera on the basis of breeding systems (Renner and Ricklefs, 1995
). Corner (1985)
viewed the evolution of functional dioecy as irreversible. Contrary to expectations based on taxonomic evidence, shifts from dioecy to monoecy in angiosperms may be more common than the reverse (Weiblen, Oyama, and Donoghue, 2000
).
Ficus is unique in that functional dioecy results from genetic factors controlling floral development combined with the impact of pollinator larvae on seed maturation (Storey, 1975
). Evidence from selection models suggests that pathways to dioecy in flowering plants may involve a gynodioecious intermediate step through the evolution of male sterility (Charlesworth and Charlesworth, 1978
), and agents of selection favoring functional dioecy in Ficus have been suggested (Kerdelhue and Rasplus, 1996
). However, there is not yet a compelling explanation for the loss of functional dioecy. It would seem that a mutation for increased ovipositor length enabling successful egg-laying in seed figs would favor a reversal to monoecy. However, the absence of pollen in seed figs prevents any offspring of a long-ovipositor mutant from founding an F2 generation. This may be precisely why the genes for style length and male sterility are linked in F. carica (Storey, 1975
). Could reversal to monoecy in functionally dioecious lineages involve the disruption of these linked loci? The possibility of pollinator behavior and morphology driving the evolution of fig breeding systems in different directions is an interesting area for future investigation.
Congruence with pollinator classification
Agreement between fig and pollinator classifications has provided a basis for speculation on the extent of coevolution between Ficus and the Agaonidae (Ramirez, 1974
; Wiebes, 1979
; Corner, 1985
). The close correspondence of fig and pollinator taxonomy could be interpreted as direct evidence for cospeciation, but congruence could also be a taxonomic artifact. Artificial agreement could arise if information from one group contributed to the classification of the other. Corner classified most of Ficus without knowledge of fig wasp taxonomy (but see pp. 395–396 in Corner, 1962
). On the other hand, Wiebes (1994a)
admitted the influence of the botanical classification in his concepts of pollinator species and genera. Phylogenetic relationships provide valuable information for identifying instances of conflict and congruence in the taxonomy of the associated groups. Phylogenetic information can then be used to evaluate whether particular cases reflect evolutionary events or taxonomic artifacts.
Agreement between fig and pollinator classification is generally supported by phylogenetic analyses based on ITS sequences and fig morphology (Fig. 9). Remarkably, there was no homoplasy in the associations of pollinator genera (CI = 1.00). Seven out of 12 pollinating genera were each associated with well-supported clades of host figs (>50% bootstrap; Fig. 6). Five of these clades represent taxonomic groups including: (1) Blastophaga-pollinated sect. Ficus, (2) Tetrapus-pollinated sect. Pharmacosycea, (3) Lipporhopalum-pollinated subsect. Paleomorphe, (4) Pleistodontes-pollinated sect. Malvanthera, and (5) Platyscapa-pollinated sect. Urostigma. In addition, a clade including sects. Kalosyce and Rhizocladus is pollinated by Wiebesia. One of the major clades of functionally dioecious figs is pollinated by Ceratosolen, corresponding subg. Sycomorus sensu Ramirez (1977)
, as designated on the basis of pollinator associations. However, some genera were associated with paraphyletic groupings of Ficus. Paraphyletic sect. Oreosycea, for example, is pollinated by Dolichoris. Also, Kradibia-pollinated sect. Sycidium is paraphyletic with respect to the Lipporhopalum-pollinated species.
|
. In this instance the conflict between fig and pollinator taxonomy is reconciled upon consideration of phylogenetic relationships. Corner (1960a)
related F. pungens to Kradibia-pollinated subsect. Sycidium, while Wiebes (1963)
placed the pollinator of F. pungens under Ceratosolen. Corner (1960a)
placed F. pungens in sect. Sycidium on the sole basis of having free tepals. However, in light of evidence from fig morphology, ITS sequences, and pollinator relationships, it is clear that F. pungens is a member of the Ceratosolen-pollinated clade.
Another interesting case concerns the placement of F. semivestita. Corner (1960b)
described the species under sect. Neomorphe based on leaves, gall figs, and growth form. In spite of similarity in buttresses, girth, and height, F. semivestita is the only member of sect. Neomorphe with axillary figs. However, Corner's original description was incomplete because seeds were unknown at the time. Recent seed collections provided additional characters supporting the placement of F. semivestita in sect. Adenosperma, including the presence of a gynobasic style and auriculiform achenes. Wiebes (1963)
suggested a close relationship between Ceratosolen grandii, the pollinator of F. semivestita, and C. appendiculatus, the pollinator of F. variegata (sect. Neomorphe). This affinity, however, was based on two homoplasious morphological characters, and phylogenetic relationships inferred from mitochondrial genes indicate that C. grandii is more closely related to pollinators of sect. Adenosperma than to C. appendiculatus (G. Weiblen, unpublished data). The results once again suggest that the classification of functionally dioecious figs can be improved based on phylogenetic analyses.
The overall congruence between fig and pollinator classifications is striking (Fig. 9), although fig phylogeny alone cannot distinguish between taxonomic artifacts and coevolutionary processes as explanations for congruent patterns. Reciprocal phylogenetic studies are needed to address this point, given that recent classification of pollinators was not independent of fig taxonomy (Wiebes, 1994a
). However, it can be concluded from phylogenetic analyses of nuclear ribosomal DNA sequences and morphology that the traditional classification of functionally dioecious figs is not supported. Subgenus Ficus is not monophyletic, and multiple reversals to monoecy have occurred within a functionally dioecious clade. Studying the evolution of fig breeding systems in relation to pollinators using a phylogenetic approach may further improve our understanding of fig/pollinator coevolution.
|
FOOTNOTES |
|---|
2 Current address: Department of Zoology, 203 Natural Sciences Building, Michigan State University, East Lansing, Michigan 48824 USA.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Barrett, M., M. J. Donoghue, and E. Sober. 1991 Against consensus. Systematic Zoology 40: 486–493[CrossRef]
Beck, N. G., and E. M. Lord. 1988 Breeding system in Ficus carica, the common fig. I. Floral diversity. American Journal of Botany 75: 1904–1912[CrossRef][ISI]
Berg, C. C. 1989 Classification and distribution of Ficus. Experientia 45: 605–611
———. 1990a Differentiation of flowers and inflorescences of Urticales in relation to their protection against breeding insects and pollination. Sommerfeltia 11: 13–34
———. 1990b Reproduction and evolution of Ficus (Moraceae): Traits connected with the adequate rearing of pollinators. Memoirs of the New York Botanical Garden 55: 169–185
Bremer, K. 1988 The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795–803[CrossRef][ISI]
Bronstein, J. 1992 Seed predators as mutualists: ecology and evolution of the fig/pollinator interaction. In E. A. Bernays [ed.], Insect-plant interactions, 1–47. CRC Press, Boca Raton, Florida, USA
Brues, C. T. 1910 The parasitic Hymenoptera of the Tertiary of Florissant, Colorado. Bulletin of the Museum of Comparative Zoology at Harvard College 54: 1–125
Buckler, E. S., A. Ippolito, and T. P. Holtsford. 1997 The evolution of ribosomal DNA: divergent paralogues and phylogenetic implications. Genetics 145: 821–832[Abstract]
Bull, J. J., J. P. Huelsenbeck, C. W. Cunningham, D. L. Swofford, and P. J. Waddell. 1993 Partitioning and combining data in phylogenetic analysis. Systematic Biology 42: 384–397[CrossRef]
Campbell, C. S., M. F. Wojciechowski, B. G. Baldwin, L. A. Alice, and M. J. Donoghue. 1997 Persistent nuclear ribosomal DNA sequence polymorphism in the Amelanchier agamic complex (Rosaceae). Molecular Biology and Evolution 14: 81–90[Abstract]
Charlesworth, B., and D. Charlesworth. 1978 A model for the evolution of dioecy and gynodioecy. American Naturalist 112: 975–997[CrossRef][ISI]
Condit, I. 1964 Cytological studies in the genus Ficus. III. Chromosome numbers in sixty-two species. Madroño 17: 153–155
Corlett, R. T. 1993 Sexual dimorphism in the reproductive phenology of Ficus grossularioides Burm. f. in Singapore. Malayan Nature Journal 47: 149–155
Corner, E. J. H. 1933 A revision of the Malayan species of Ficus: Coveliia and Neomorphe. Malayan Branch Royal Asiatic Society 11: 1–65
———. 1955 Revision of Ficus. Flora Malesiana Bulletin 11: 428–430
———. 1958 An introduction to the distribution of Ficus. Reinwardtia 4: 325–355
———. 1960a Taxonomic notes on Ficus Linn., Asia and Australasia. I–IV. Gardens' Bulletin Singapore 17: 368–485
———. 1960b Taxonomic notes on Ficus Linn., Asia and Australasia. V–VI. Gardens' Bulletin Singapore 18: 1–69
———. 1961 Taxonomic notes on Ficus Linn., Asia and Australasia: Addendum. Gardens' Bulletin Singapore 18: 83–99
———. 1962 Taxonomic notes on Ficus Linn., Asia and Australasia: Addendum II. Gardens' Bulletin Singapore 19: 385–401
———. 1965 Check-list of Ficus in Asia and Australasia with keys to identification. Gardens' Bulletin Singapore 21: 1–186
———. 1967 Ficus in the Solomon Islands and its bearing on the Post-Jurassic history of Melanesia. Proceedings of the Royal Society of London, Series B 253: 23–159
———. 1969 Ficus sect. Adenosperma. Philosophical Transactions of the Royal Society of London, Series B 256: 319–355
———. 1970a Ficus subgen. Ficus. Two rare and primitive pachycaul species. Philosophical Transactions of the Royal Society of London, Series B 259: 353–381[CrossRef]
———. 1970b Ficus subgen. Pharmacosycea with reference to the species of New Caledonia. Philosophical Transactions of the Royal Society of London, Series B 259: 383–433[CrossRef]
———. 1976 The climbing species of Ficus: derivation and evolution. Philosophical Transactions of the Royal Society of London, Series B 273: 359–386[CrossRef]
———. 1978 Ficus dammaropsis Bl. and the pedunculate species of Ficus subgen. Sycocarpus. Philosophical Transactions of the Royal Society of London, Series B 281: 373–406
———. 1985 Ficus (Moraceae) and Hymenoptera (Chalcidoidea): figs and their pollinators. Biological Journal of the Linnean Society 25: 187–195[CrossRef]
de Queiroz, K. 1996 Including the characters of interest during tree reconstruction and the problems of circularity and bias in studies of character evolution. American Naturalist 148: 700–708[CrossRef][ISI]
———, M. J. Donoghue, and J. Kim. 1995 Separate versus combined analysis of phylogenetic evidence. Annual Review of Ecology and Systematics 26: 657–681[ISI]
Donoghue, M. J., and D. A. Ackerly. 1996 Phylogenetic uncertainties and sensitivity analyses in comparative biology. Philosophical Transactions of the Royal Society of London, Series B 351: 1241–1249[CrossRef]
———, R. G. Olmstead, J. F. Smith, and J. D. Palmer. 1992 Phylogenetic relationships of Dipsacales based on rbcL sequences. Annals of the Missouri Botanical Garden 79: 333–345
———, and M. J. Sanderson. 1992 The suitability of molecular and morphological evidence in reconstructing plant phylogeny. In P. S. Soltis, D. E. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants, 340–368. Chapman and Hall, New York, New York, USA
Doyle, J. J., and J. L. Doyle. 1987 A rapid DNA isolation procedure for small amounts of fresh leaf tissue. Phytochemical Bulletin 19: 11–15
Farris, J. S., M. Kallersjo, A. G. Kluge, and C. Bult. 1994 Testing significance of incongruence. Cladistics 10: 315–319[CrossRef][ISI]
Felsenstein, J. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791[CrossRef][ISI]
Galil, J. 1973 Pollination in dioecious figs: pollination of Ficus fistulosa by Ceratosolen hewitti. Gardens' Bulletin Singapore 26: 303–311
Ganeshaiah, K. N., P. Kathuria, R. U. Shaanker, and R. Vasudeva. 1995 Evolution of style-length variability in figs and optimization of ovipositor length in their pollinator wasps: a coevolutionary model. Journal of Genetics 74: 25–39[ISI]
Gasparrini, G. 1844 Nova genera quae super nonnullis Fici speciebus struebat, Naples
Grafen, A., and H. C. J. Godfray. 1991 Vicarious selection explains some paradoxes in dioecious fig-pollinator systems. Proceedings of the Royal Society of London, Series B 245: 73–75[CrossRef]
Herre, E. A., C. A. Machado, E. Bermingham, J. D. Nason, D. M. Windsor, S. S. McCafferty, W. V. Houten, and K. Bachmann. 1996 Molecular phylogenies of figs and their pollinator wasps. Journal of Biogeography 23: 521–530[CrossRef][ISI]
Hillis, D. M., and J. J. Bull. 1993 An empirical test of bootstrapping as a method for assessing confidence in phylogenetic analysis. Systematic Biology 42: 182–192[CrossRef][ISI]
Hossaert-McKey, M., M. Gibernau, and J. E. Frey. 1994 Chemosensory attraction of fig wasps to substances produced by receptive figs. Entomologia Experimentalis et Applicata 70: 185–191[CrossRef][ISI]
Janzen, D. H. 1979 How to be a fig. Annual Review of Ecology and Systematics 10: 13–51
Kelley, L. M. 1998 Phylogenetic relationships in Asarum (Aristolochiaceae) based on morphology and ITS sequences. American Journal of Botany 85: 1454–1467
Kerdelhue, C., M. E. Hochberg, and J. Y. Rasplus. 1997 Active pollination of Ficus sur by two sympatric fig wasp species in West Africa. Biotropica 29: 69–75
———, and J. Y. Rasplus. 1996 The evolution of dioecy among Ficus (Moraceae): an alternative hypothesis involving non-pollinating fig wasp pressure on the fig-pollinator mutualism. Oikos 77: 163–166[CrossRef][ISI]
King, G. 1887 On some new species of Ficus from New Guinea. Journal of the Asiatic Society of Bengal 55: 394–411
Kjellberg, F., P. H. Gouyon, M. Ibrahim, M. Raymond, and G. Valdeyron. 1987 The stability of the symbiosis between dioecious figs and their pollinators: a study of Ficus carica L. and Blastophaga psenes L. Evolution 41: 693–704[CrossRef][ISI]
Kluge, A. J. 1989 A concern for evidence and a phylogenetic hypothesis of relationships among Epicrates (Boidae, Serpentes). Systematics Zoology 38: 7–25
